|
Activity Number:
|
384
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 15, 2002 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
General Methodology
|
| Abstract - #301135 |
|
Title:
|
The Effect of Missing Data on Linear Filters
|
|
Author(s):
|
Alexia Iasonos*+
|
|
Affiliation(s):
|
State University of New York, Albany
|
|
Address:
|
One University Place, Rensselaer, New York, 12144-3456, USA
|
|
Keywords:
|
missing values (weekend effect) ; frequency domain ; KZ filter ; transfer function
|
|
Abstract:
|
Filters are widely used in time series analysis to decompose a series into permanent and temporary components. We used the Kolmogorov-Zurbenko filter in conjunction with frequency domain techniques to separate the different components of historical financial data (Iasonos and Zurbenko, 2001). As it is the case with other data, missing values can disturb the frequency analysis tool commonly used, the periodogram. It is thus essential to ensure that the transfer function of the filter is not altered by the presence of missing observations (weekend effect). We show that the actual transfer function of a filter, when missing values are present, converges in mean square to the theoretical transfer function when complete data are available. The distribution of the difference of the two transfer functions can be approximated by the sum of two squared normally distributed variables and asymptotically follows Chi-square distribution with two degrees of freedom. The results show that the difference between the theoretical and actual transfer functions vanishes when the missing rate approaches zero. We validate our results geometrically and by simulation.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2002 program |
